Google Ads Audit Checklist You Can Run in an Hour

Why most Google Ads audit checklists never get finished

Because they are written to be handed over, not run. A 64-point scored checklist makes an excellent pitch document — it proves the auditor looked at everything — and a terrible weekly habit. Nobody gets to item 19 twice.

The deeper problem is that those lists mix two things that move at completely different speeds. Checking whether your conversion action is set to "Primary", whether your campaigns share a budget, whether your sitelinks are approved — those are structural. They change two or three times a year. You can audit them in January and mostly be right in June.

Search terms are not like that. Every week, Google matches your keywords and your Performance Max assets to queries nobody typed last week. An account that was clean on Monday has new junk in it by Friday. Auditing that once a year and scoring it out of 64 is like sweeping a shop floor in January and declaring the shop clean.

So split the job. Keep your structural checklist, run it quarterly, use whichever free template you like. Then run this separately, and run it often. It is the half where the money actually leaks, and it fits in an hour.

The hour, in five passes

Set a timer for each pass. The timer is not a gimmick — it stops you doing the thing that kills these audits, which is going down a rabbit hole on term number four and never reaching term number ninety.

That is sixty minutes. If you overrun, you overran on pass 2, and pass 2 is the one you were supposed to do fast and rough.

Pass 1 — Pull 90 days of search terms sorted by cost

One report, three filters. Search terms report, last 90 days, conversions = 0, sorted by cost descending. That's it. Everything else you could pull is a distraction for this hour.

If you work in the interface: Campaigns → Insights and reports → Search terms. Add the columns cost, clicks, conversions and conversion value. If you pull data through the API or a script, the resource is campaign_search_term_view, and it covers Search, Shopping and Performance Max in one place.

Why 90 days and not 30? Because a term that spent $12 in a month looks like nothing, and the same term at $36 a quarter looks like what it is. Low-cost, high-frequency waste is invisible in short windows. Why not 12 months? Because your keywords, your products and your landing pages have changed, and you will spend the hour arguing with yourself about terms that stopped showing in March.

Now set a threshold, so you know where to stop reading. The cleanest one is your target CPA. If your target cost per acquisition is $80, then a search term that has spent $240 with zero conversions has had three full chances and taken none of them. Everything above that line goes on the list. Everything below it can wait for next month's pass — it will still be there, and it will have spent more by then, which makes the decision easier.

In a Search and Shopping account spending $8,000 a month, that 90-day report typically looks something like this: a few hundred terms with clicks, of which maybe 180 have zero conversions and account for around $3,100. Of those 180, perhaps 60 are unmistakably wrong — $1,200 a quarter, $4,800 a year, for traffic that was never going to buy. Your numbers will differ. The shape rarely does. Our longer walkthrough of how to find wasted ad spend in Google Ads goes deeper on the maths.

One thing to know before you trust the total: the costs in the search terms report will not add up to your campaign spend. Google withholds terms that too few people searched, and those withheld terms still cost you money. The gap is real and it is not a bug in your export — we cover what disappears and why in the search terms report and what it hides.

Pass 2 — Four buckets, not one decision per term

The reason term-by-term judgement takes three hours is that every term becomes its own small debate. Bucket first. Decide second. You are sorting, not deliberating.

Bucket 1 — wrong intent. The searcher wants information, a job, a salary figure, a DIY method, or a free version. For a plumbing company: plumber salary, how to unblock a sink yourself, plumbing apprenticeship near me, free plumbing advice forum. These people are not one landing page away from buying. They are not customers. Job-related traffic in particular is a large, recurring bucket in service businesses — there is a ready-made list in job seekers clicking Google Ads.

Bucket 2 — wrong product or service. You sell commercial espresso machines and you paid for nespresso pods. You fit boilers and you paid for air conditioning installation. The intent is commercial, the money is real, the product is not yours.

Bucket 3 — wrong place or wrong scale. You serve three US states and you paid for clicks on emergency plumber sydney. You do B2B minimum order 500 units and you paid for buy 1 phone case.

Bucket 4 — right person, no conversion yet. emergency plumber near me, $310 spent, zero conversions. This one is not a negative keyword. It is either a tracking problem, a phone-call problem, a landing page problem or a price problem, and blocking it would be the most expensive mistake you make this hour. Write it on a separate line and leave it alone. If bucket 4 is most of your list, stop the audit — your issue is upstream, and the order in which to check things when Google Ads spends with no conversions is a better use of the next hour than any negative keyword.

Buckets 1, 2 and 3 get negatives. Bucket 4 gets a note.

Competitor names are the one genuinely contested category, and they do not belong in any of the four buckets by default. Whether [competitor] alternative is waste or your best-converting segment depends on your margins and your product, and we argue both sides in competitor searches, block or keep. During the hour, put them aside. Deciding competitor policy is a separate, slower conversation.

While you sort, you will notice patterns rather than individual terms: everything containing salary, everything containing jobs, everything containing free, everything containing a city you don't serve. Patterns are worth more than terms, because a pattern blocks the twelve variants you haven't seen yet. If you're wondering why the same shapes keep reappearing month after month, why Google Ads shows irrelevant search terms explains the matching behaviour behind it.

Pass 3 — Performance Max and Shopping, on their own terms

Ten minutes, because the rules are different and applying Search habits here will silently fail.

Performance Max accepts up to 10,000 negative keywords at campaign level. It does not accept shared negative keyword lists. That single fact reshapes the pass: the tidy account-level list you maintain for Search does not reach your PMax campaigns at all. If you run four PMax campaigns, the same exclusion has to be written four times. Ten thousand sounds infinite until you are maintaining it by hand across several campaigns — we look at what actually fills that space in negative keywords in Performance Max and the 10,000 cap.

PMax waste also hides in places a Search-shaped audit doesn't look. Brand queries absorbed by the campaign, search themes drifting, asset groups pulling traffic you'd never bid on deliberately — where Performance Max wasted spend hides is the map for that ten minutes.

Standard Shopping is friendlier: it takes negative keywords and shared lists, the same way Search does. But the way a negative interacts with a product feed is not the way it interacts with a keyword, and a broad negative that seemed harmless can pull a whole product group out of the auction. Shopping campaign negative keywords work differently covers the traps.

Display, Video and App campaigns report no search terms at all. There is nothing to audit here in this hour, and no negative keyword will fix them. If those campaigns are burning money, that is a placement and audience problem, on a different day.

Pass 4 — Write the negatives so you can undo them

Fifteen minutes. Two decisions per negative: which match type, and which level.

Match type. Default to phrase. Use exact when the term is fine in most contexts and wrong in one — blocking exact plumber while keeping emergency plumber. Be wary of broad negatives. Broad, phrase and exact are the three negative match types, and broad negatives behave unpredictably across match types in a way that surprises people who assume negatives mirror positives. They do not. Negative keyword match types don't work like positives is worth reading once before you write a hundred of them.

The practical rule for the hour: if you are about to add a broad negative containing a word that also appears in a term you want to keep, stop and write it as phrase instead. salary as a broad negative is fine. free as a broad negative will take free delivery and free quote with it.

Level. Account-level or shared-list negatives for things that are wrong everywhere — job searches, a city you never serve, a product you don't sell. Campaign-level for things that are wrong in one campaign and right in another. Ad-group level almost never, during an hour audit: it is the level you forget exists when you come back in six weeks and wonder why a term is still blocked.

One caution before you paste 200 lines in: a large negative list is not free. It has a maintenance cost, it can collide with terms you later want, and past a certain size nobody on the team knows what's in it. How many negative keywords are too many puts numbers on that. If your list is already enormous, pass 4 should include deleting a few as well as adding.

Pass 5 — The log, which is the part checklists never have

Ten minutes, one row per decision: date, term or pattern, match type, level, reason, campaign. A spreadsheet is fine.

This is the difference between an audit and a habit. Without a log, three things happen. Next month you re-litigate free from scratch, because you can't remember whether you decided to block it or keep it. When conversions dip in six weeks, you have no list of what changed and you end up deleting negatives at random. And when someone else takes over the account, every exclusion looks arbitrary, so they either keep all of them out of fear or delete all of them out of frustration.

The reason column matters more than it looks. "Job seeker" and "we don't sell this" and "wrong country" age differently. When you open a second warehouse in Texas, you want to find every negative whose reason was "wrong country" in ten seconds. You will never find them by reading 400 keyword strings.

If you'd rather not keep the log by hand, that is roughly what PPCleaner is: every block it makes is recorded with the term, the campaign, the cost that triggered it and the date, and every block is reversible in one click for 90 days. It also never blocks a search term that converted in the last 90 days, which removes the single most expensive error in this whole hour. What it will not do is tell you which bucket-4 terms need a better landing page. That judgement stays with you.

What one hour cannot tell you

Four honest limits, worth knowing before you present the findings to anyone.

Your totals will not reconcile. Google withholds low-volume search terms. The spend you can see and judge is a subset of the spend that happened. In the accounts we audit the visible portion is usually most of it but rarely all of it, and the gap varies a lot by account. Don't promise a client that you've accounted for every dollar.

"Zero conversions" is only as good as your tracking. If your form thank-you page fires twice, if phone calls aren't tracked, if offline sales never come back into the account, then bucket 4 is full of terms that did convert and you cannot see it. A waste audit run on broken tracking will confidently block your best traffic. Check tracking first, every time, and if you can't verify it in ten minutes, postpone the audit.

Conversion lag eats the end of your window. If your typical sale closes 21 days after the click, the last three weeks of your 90-day report are partially unread. Terms from that period will look worse than they are. Either exclude the most recent lag-length from your judgement, or raise your cost threshold for recent terms.

This hour audits one thing. It does not check bids, budgets, ad copy, landing pages, audience settings, bidding strategy or account structure. Nothing in these five passes will fix a campaign whose real problem is a 9-second load time. Waste auditing is necessary and it is narrow. Say so out loud, especially when someone is paying you for the hour.

Running it again is what makes it work

A single audit clears what accumulated. It does not stop accumulation. The terms you blocked today are gone; the ones Google will match you to next Tuesday are not in your report yet.

So the real question is cadence, not completeness. For most accounts under $10,000 a month, monthly is enough, and weekly is enough for anything above that or anything seasonal. We go through the trade-offs of each frequency in how often you should check search terms. The second run is always faster than the first — thirty minutes, not sixty — because the patterns are already blocked and the log tells you what you already decided.

Twelve hours a year, spread across twelve sittings, is a real cost. Whether that's better spent in-house or handed to someone else depends on the account size and on what your own hour is worth, which is the arithmetic in agency or do it yourself. And if the answer is "neither, I want this to happen without me", that is exactly the gap PPCleaner fills — the same five passes, run nightly, on Search, Shopping and Performance Max, with 90 days of one-click undo behind every decision. It touches nothing else: not budgets, not bids, not ads, not keywords.